The Dynamics of Interhemispheric Collaboration and Hemispheric Control

36, 193–208 (1998) BR970952

BRAIN AND COGNITION ARTICLE NO.

The Dynamics of Interhemispheric Collaboration and Hemispheric Control Jacqueline Liederman Brain, Behavior, and Cognition Program, Boston University This article reviews recent models of how the two hemispheres collaborate to facilitate efficient information processing. The strengths and weaknesses of behavioral techniques and structural and functional neuroimaging techniques are considered as they apply to these problems. The role of the corpus callosum as an equilibrator of activation between the two hemispheres is discussed and linked with its role as a dynamic modulator for the mobilization of hemispheric resources. Issues related to metacontrol and dynamic fluctuations in hemispheric control are also considered.  1998 Academic Press

INTRODUCTION

This article is written in honor of M. Philip Bryden, who died suddenly this past summer while attending the Montreal Conference TENNET: Theoretical and Experimental Neuropsychology (Neuropsychologie Expe´rimentale et The´orique). At TENNET he was the discussant at a session devoted to interhemispheric interaction, the papers of which are published in this volume. Hence, this commentary on those papers serves not only to commemorate his role in that event, but also to celebrate his innumerable contributions to the field of interhemispheric interaction. Variables that govern hemispheric dominance and interhemispheric interaction were very much on the mind of Dr. Bryden at the time of his death. One has only to look at the great number of research abstracts on this topic that were submitted from his laboratory to the International Neuropsychological Society that summer. For example, Asbjørnsen, Bryden, and Ofte (1997) studied voluntary and automatic shifts of attention toward a side of space, and Asbjørnsen and Bryden (1997) invented an Attention Shift Index. This Index computed the extent to which an individual’s attention can be shifted by cues from one ear to the other during dichotic listening. The unique aspect Address correspondence and reprint requests to Jacqueline Liederman, Ph.D., Director of the Brain, Behavior, and Cognition Program, Boston University, 64 Cummington Street, Boston, MA 02215. 193 0278-2626/98 $25.00 Copyright  1998 by Academic Press All rights of reproduction in any form reserved.

194

JACQUELINE LIEDERMAN

of the Index was that it was designed with statistical properties that permitted one to test for significance of attentional shifts within individual subjects. Other individuals associated with Dr. Bryden investigated individual and developmental differences in cerebral specialization and so-called ‘‘complementarity’’ (Boucher & Bryden, 1997; Crebolder & Bryden, 1997; Nikelski, Grimshaw, Bryden, & Cocivera, 1997; and Obrzut, Bryden, Lange, & Bulman-Fleming, 1997). Others in his laboratory were devoted to the study of bihemispheric interactions (Keillor, Grimshaw, & Bryden, 1997). In fact, one of the three papers presented at the interhemispheric symposium at TENNET (by Dr. Grimshaw) was from Dr. Bryden’s laboratory. The research was an examination of the extent to which the two hemispheres can work in parallel without mutual interference. All three papers that were presented at the TENNET Conference attempted to define the circumstances in which bihemispheric presentation and/or processing exceeds the unihemispheric processing of input. Each author, however, approached this issue with a radically different method, set of assumptions, and theoretical explanations. Let us examine each in turn. METHODS Each of the authors departs from the original method introduced by Dimond (1972) in The Double Brain. In Dimond’s original studies, differences between unihemispheric and bihemispheric processing were inferred from responses to unilateral (i.e., unihemispheric) presentation vs bilateral presentation (wherein each hemisphere received only half of the stimuli). Division of inputs between the hemispheres was often advantageous and is referred to as the ‘‘bilateral advantage.’’ Instead of dividing the inputs between the hemispheres and comparing performance in that situation to that resulting from unilateral stimulation, Banich (1998), Grimshaw (1998), and Hellige, Taylor, Lesmes, and Peterson (1998) use different methodological approaches to address the question. For example, in Banich’s (1998) paradigm, there are always three stimuli in a visual display, arranged in a skewed V-shape. The stimuli at the top of the V are strongly lateralized, one to each side of space. The third stimulus is at the bottom of the V, lateralized to one of the sides but closer to the midline than the top two stimuli. For each trial, two stimuli are presented to one hemisphere and one to the other. On unilateral trials, the two items that are relevant to the task are presented to the same side (at the top and base of the V). On bilateral trials, the items that are relevant to the task are presented to opposite hemispheres (one at the top of the V on one side and one at the base of the V on the other side). Note that the subjects are not given any cues as to whether the inputs on one side or the other would be relevant. The ‘‘relevant’’ items during bilateral trials are somewhat further apart than they are during unilateral trials, but other than that the trials are perceptually identical. The subject must begin to decode all three inputs until a match is found, irrespective of whether the trial is unilateral or bilateral. The question is which arrangement is advantageous. Grimshaw (1998) employs an entirely different method for inferring whether bihemispheric processing is advantageous. She examines whether interference between conflicting aspects of a single input is reduced when the individual is able to process the two dimensions of that input in opposite hemispheres as opposed to a single hemisphere. As Grimshaw (1998) argues: ‘‘Consider the common situation in which each hemisphere performs computations on the same input, but in qualitatively different ways that reflect hemispheric specialization.’’ Thus, in Grimshaw’s (1998) paradigm, a situation is created in which, for most individuals, the left

INTERHEMISPHERIC COLLABORATION

195

hemisphere will be predominant for processing one dimension of the input whereas the right hemisphere will be predominant for processing a second dimension of that same input. Hellige et al.’s paradigm is similar to Grimshaw’s (1998) in the sense that it examines the pattern of performance that emerges when both hemispheres have equivalent access to incoming stimulus information. Hellige et al. (1998) compare performance in response to unilateral presentation of a stimulus to performance in response to the same stimulus presented redundantly to the right and left sides of space simultaneously. Thus, the bilateral advantage that they found is a result of stimulus redundancy across hemispheres and not the advantage associated with presentation of unique inputs to opposite hemispheres.

FINDINGS AND THEORETICAL IMPLICATIONS

Banich operates on the assumption that within-hemisphere computations will be intrinsically advantageous compared to dual-hemisphere computations because of the complexity of interhemispheric coordination of the products resulting from processing. Her assumptions are supported by Ringo, Doty, Demeter, and Simard (1994), who argue that some interhemispheric interactions may take as long as 100–300 ms because they travel along thin unmyelinated fibers. Banich argues, however, that as task difficulty increases beyond the capacity of the resources of a single hemisphere, or when information engenders conflict, then bihemispheric processing is advantageous. As one hemisphere’s resources or capacity is overtaxed by the processing requirements placed upon it, the other side of the brain will be recruited or activated to underwrite the processing. Banich conceptualizes this as the role of interhemispheric interaction, mediated by the corpus callosum, in dynamically modulating the processing capacity of the whole brain. Banich reviews a lovely stream of experiments from her laboratory, which have systematically demonstrated that there are many aspects of computational complexity that can shift the balance in favor of bihemispheric processing. Thus, across different modalities, she has demonstrated that performance of more computationally complex tasks is aided by interhemispheric interaction. Banich defines computational complexity ‘‘as dependent on the number and sorts of transformations, operations, or computations that must be performed on an input before a decision can be reached.’’ One of her unique contributions has been to reformulate the notion of ‘‘complexity’’ so as to be inclusive not only of tasks with more items, or more stages of processing, but also of those situations in which the attentional demands of a task are increased by the inclusion of distracting information. Thus, tasks that involve interitem interference are conceptualized by Banich as more difficult than those that do not involve interitem interference. In this manner, she can generalize that the more attention a task demands, the more likely it is that division of inputs between the hemispheres will be beneficial. This is a powerful prediction. For example, using the three-item V displays, Banich (1998) examines performance when subjects have to decide whether or not hierarchical figures match on a particular cued dimension. In some cases, the items match at

196

JACQUELINE LIEDERMAN

both the global and the local levels. Therefore, if subjects are cued to attend to the local level, the global level is not distracting. In other cases, items match on only the irrelevant dimension (i.e., global when the dimension-tobe-attended is local). In those cases, the subject has to ignore the second dimension in order to avoid making a false-positive response. When matching items are presented to different hemispheres there is better performance than when they are presented to the same hemisphere. In particular, this is the case during trials when the irrelevant dimension matches the target, and because Banich argues that these trials are computationally more complex than those in which both the relevant and irrelevant dimensions match the target, she is able to predict this outcome strictly on the basis of computational complexity. A recent series of experiments from Liederman’s laboratory (Liederman, Sohn, Thome, & Palomo, 1995; Sohn, Liederman, & Reinitz, 1996) present the first data that directly demonstrate that certain kinds of interitem interference simply do not occur when inputs are divided between the hemispheres. Many experiments are consistent with this interpretation, but these are the first to actually demonstrate that the pattern of error types is qualitatively different during uni- compared to bihemispheric presentation. The experiments utilize illusory conjunction paradigms, which require the subject to search through an array of four words (Liederman et al., 1995) or four colored letters (Sohn et al., 1996) in order to find a match to a prespecified target word (Liederman et al., 1995) or colored letter (Sohn et al., 1996). For half of the trials one of the four items matches the target; for the other half none matches. Within these nonmatching trials, subjects can make one of two kinds of false-positive errors. During the ‘‘conjunction’’ trials, both of the elements of the target are in the search display. For example, if the target were a red-colored letter ‘‘T,’’ one item in the display might be a redcolored letter ‘‘O’’ and the other item might be a green-colored letter T. To respond ‘‘match,’’ the subject must miscombine these separate items into an illusory perception of their having occurred within a single item. During the ‘‘feature’’ error trials, one of the elements of the target is presented twice within the search display. For example, if the target were a red-colored T, one item in the display might be a red-colored letter ‘‘X’’ and the other might be a red-colored letter ‘‘O’’. To respond match, the subject would have to misperceive one of those letters as a T. It is inferred that subjects are miscombining information and not just misperceiving information, when the error rate during conjunction (i.e., miscombination) trials exceeds the error rate during feature (i.e., misperception) trials. If the advantage for bihemispheric presentation arises only when computational complexity reaches a certain threshold, then misperceptions should be aided to an equal extent as are miscombinations by the division of inputs between the hemispheres. But if the critical aspect about the bihemispheric trials is that they

INTERHEMISPHERIC COLLABORATION

197

separate inputs from mutual interference, then the division of inputs between the hemispheres should specifically reduce miscombination errors. In both experiments, Liederman and her colleagues found that miscombination errors were disproportionately reduced compared to misperceptions in the bihemispheric trials. Indeed, in both experiments, feature errors (i.e., misperceptions) were held constant in the uni- and bihemispheric trials, by using error rates during feature trials to independently adjust exposure time for the uni- vs the bihemispheric trials. Indeed, in both experiments true illusory conjunctions by the definition of Treisman and Schmidt (1982) occurred only during the unihemispheric trials. Miscombination errors did not exceed misperception errors during bihemispheric trials in any of the experiments. Thus, our results seem to support the notion that division of inputs is most beneficial in the advantage it provides for reducing interitem interference. These results are compatible with Banich’s scenario that one aspect of computational complexity is the extent of interitem interference. From a historical perspective, it is interesting to note that several of the elements of Banich’s theory are consistent with the early writings of Kinsbourne and his successors. What Banich (1998) has succeeded in doing is unifying these separate ideas into a single well-constructed theory. Kinsbourne had several basic notions. The first idea was that ‘‘the forebrain commissure is more concerned with excitation–inhibition balance than with information transmission’’ (Kinsbourne, 1982, p. 415). Thus, depending upon task demands, the corpus callosum will either concentrate activation upon one hemisphere or distribute activation between hemispheres. Without such equilibration, attention will swing from one hemisphere to the other. Indeed, it has been argued that distribution of activation between cortical regions is a major role of not only interhemispheric fibers, such as those in the corpus callosum, but of long-distance, intrahemispheric cortico-cortical pathways as well (Liederman, 1995). A closely related second idea was that each of the cerebral hemispheres, rather than being absolutely specialized for just one particular function, was part of a highly interconnected ‘‘multipurpose’’ computational space (Kinsbourne & Hicks, 1978). They argued that as the capacity of a specialized processor became overtaxed, associated regions would be recruited to underwrite the required processing. In a later paper, Kinsbourne (1987) reviewed data from paradigms that required two tasks to be performed at the same time, so-called dual-task paradigms. For example, the first task might involve input to a single hemisphere (e.g., tachistoscopic display to one visual field, or monaural presentation) or it might require a motor output from a single hemisphere (e.g., tapping with one hand). The second task was usually presented to both hemispheres but was presumed to require the specialized processing of one hemisphere more than the other (e.g., memorizing words from lists of varying lengths). As the number of words required to be held in

198

JACQUELINE LIEDERMAN

memory increased, performance on the primary task—tapping with the right hand—would diminish. The kind of data that inspired Kinsbourne’s hypotheses was the decline in left-hand performance as task difficulty increased. At low levels of task difficulty, the verbal task did not interfere at all with the right hemisphere’s control of the left hand. However, once difficulty for the secondary verbal-memory task passed a certain threshold, performance of the primary task by the left hand was significantly impaired. He reasoned that when the processing capacity of the left hemisphere was exceeded (e.g., by increasing the number of words required to be remembered), the right hemisphere was recruited to provide additional capacity, and this interfered with the right hemisphere’s ability to optimally control left-hand output. Note that at this point, Kinsbourne emphasized how interference became bilateral with increasing task difficulty. The fact that cerebral functional organization was flexibly altered to allow the right hemisphere to participate in processing was only secondarily emphasized. The third idea was the notion that ‘‘the economy in separating control centers in functional cerebral space resides not in the expectation of choice, or even of doing two different things at the same time, but in the utility of keeping apart and interference-free two distinct mental operations that represent concurrent contributions to the same total activity’’ (Kinsbourne, 1982, p. 417). Kinsbourne and Hicks (1978) argued that regions in opposite hemispheres were capable of insulating conflicting processes better than regions within the same hemisphere because the relative lack of interhemispheric (as compared to intrahemispheric) fibers rendered the two hemispheres ‘‘functionally distant.’’ Years later, Merola and Liederman (1987) used a dual-task tachistoscopic paradigm that required subjects to perform two incompatible operations at the same time, namely identifying upright and inverted letters. They demonstrated that an advantage for the division of inputs between the hemispheres, in comparison to single-hemisphere input, emerged only when task difficulty was increased by doubling the number of stimuli per trial from two to four. Similarly, Norman, Jeeves, Milne, and Ludwig (1992) found that increases in task difficulty in terms of number of items increased the bilateral advantage. Thus Banich has integrated these disparate ideas into the simple prediction, that as computational complexity increases, the advantage of the distribution of processing across the hemispheres should increase. Her behavioral data have been supportive of this prediction. Beyond these behavioral paradigms, one might propose that modern neuroimaging methods (such as functional magnetic resonance imaging, fMRI) might be used to visualize how the balance of hemispheric activation shifts in response to increases in computational complexity. Banich (1998) would predict a shift from predominantly single-hemisphere activation to bilateral activation as task difficulty increased. Unfortunately, with today’s methods, it is still not possible to reliably infer differential degrees of activation above

INTERHEMISPHERIC COLLABORATION

199

baseline across regions by these imaging techniques. This is because the degree of activation of a region of interest over baseline is often on the order of only 3%. The difficulty in the use of fMRI for such investigations is well illustrated by a recent controversy that appeared in Science. Just, Carpenter, Keller, Eddy, and Thulborn (1996) presented data that they claim show that, as the linguistic complexity of sentences increases, the brain recruits larger regions of neural tissue. Just et al. measured neural activation by means of fMRI and found that the number of voxels showing activation above baseline increased significantly across the three levels of sentence complexity. According to Just et al., this demonstrated two things: (1) that more neural tissue was recruited in response to increased computational demands and (2) that there ‘‘cannot be a static cartography of brain anatomy.’’ The controversy, initiated by a letter to the editor by Rapp and McCloskey (1997), centered over whether there had been actual recruitment of new tissue or merely increased activation of the same regions previously activated by the less complex task. Indeed, Rapp and McCloskey (1997) argued that recruitment of new tissue may be an illusion due to a statistical artifact. Thus, at low-task complexity many voxels may be only marginally active compared to baseline. At high-task complexity many more voxels may be classified as significantly active compared to baseline. According to Rapp and McCloskey (1997), this would not indicate recruitment of tissue; it would only indicate that task complexity led to greater activation of the same tissue. Nor was there any evidence in the Just et al. (1996) paradigm that unilateral activation shifted to bilateral activation as task complexity increased. Even at the lowest level of sentence complexity, there was bilateral activation of Broca’s area and of its right-sided homologue. As sentence complexity increased, signal intensity increased on both the right and the left sides. This would be an example of increased activation rather than recruitment. Hence, the notion that as task complexity increases, bilaterality of activation increases may, at this time, be hard to evaluate except by the elegant but indirect behavioral methods that Banich (1998) has already employed. Grimshaw’s (1998) work attempts to delineate a different advantage of bihemispheric processing. Based on the notion that potentially conflicting processes are better performed in opposite hemispheres (Kinsbourne & Hicks, 1978; Kinsbourne, 1982; Liederman, 1986; Liederman, Merola, & Hoffman, 1986; Liederman & Meehan, 1986; Merola & Liederman, 1985, 1987, 1990), Grimshaw predicts that individuals who process both dimensions of the stimulus in a single hemisphere will experience greater interference than those who process each dimension in opposite hemispheres. Grimshaw’s paradigm is beautifully designed. Building upon previous work in Dr. Bryden’s laboratory, which demonstrated that the right hemisphere is specialized for the analysis of prosody (Ley & Bryden, 1982; Bryden &

200

JACQUELINE LIEDERMAN

MacRae, 1988; Bulman-Fleming & Bryden, 1994), Grimshaw devised a set of stimuli that have conflicting linguistic and prosodic processing dimensions. The task required decoding a message wherein the content of the message and the tone of voice of the speaker are in conflict. Subjects were presented with words referring to an emotion (mad, sad, glad) or a neutral subject (fad) conveyed in either a compatible or incompatible tone of voice (mad, sad, glad, or neutral voice). During different blocks of trials, the subjects were required to identify either the word or the emotional tone of the voice. Answers were indicated by pressing one of four buttons that subjects were trained to think of as referring to one of the four emotional words/ emotional voices. In a different part of the experiment the inputs were lateralized and the subjects were categorized in terms of whether they manifest a right- or leftear advantage for the individual tasks of word identification or emotional tone identification. The hypothesis was that in individuals with hemispheric complementarity of specialization, within whom one hemisphere is specialized for language and the other for prosody and the interpretation of emotion, there would be a greater capacity for one dimension to be processed without interference from the other dimension compared to individuals who chose to process both dimensions of the input in the same hemisphere. Grimshaw’s task worked perfectly in that significant two-way interference effects were obtained between meaning and tone. Identification of the emotion word was slowed when the word was conveyed in a tone of voice connoting a different emotion. Similarly, identification of the emotional tone of the speaker was slowed when the emotion word named by the speaker referred to a different emotion. No reduction in either of these interference effects was found in those individuals who were classified as performing the linguistic and prosodic tasks in opposite hemispheres as opposed to the same hemisphere. Therefore, the hypothesis that individuals with complementary hemispheric specialization would show the least intertask interference was not confirmed. Despite the elegance of the design, there are several possible reasons (in addition to the power problem, which Grimshaw acknowledges) why the intended results may not have been obtained. Grimshaw (1998) suggests that the problem is that the task requires the subject to actually label one dimension of the input. Both hemispheres need to resolve their response conflict before a single finger can be selected to indicate the identification response. It is known, however, that the division of inputs between the hemispheres does not seem to reduce interference at the response-selection stage. Pashler and O’Brien (1993) demonstrated that as stimulus onset asynchrony was shortened between two successively presented tasks, the delay in subjects’ responding was not affected by whether the tasks were presented to a single hemisphere or to separate hemispheres. Because interference in this ‘‘psychological refractory’’ paradigm is typically attributed to competition be-

INTERHEMISPHERIC COLLABORATION

201

tween responses, division of inputs between the hemispheres did not reduce interference based on response selection. That is certainly one possible explanation for Grimshaw’s (1998) experiment not having yielded the expected demonstration that processing of the competing dimensions of the auditory signal would be more efficient in individuals who tend to perform that processing in separate hemispheres. Hellige (1995) provides another line of reasoning that could be used to explain Grimshaw’s (1998) (negative effects) results. Hellige attempts to explain why, in his own data, bilaterally redundant presentations of consonant–vowel–consonant stimuli do not result in error patterns suggestive of left-hemisphere metacontrol of the task. He argues that the modes of processing favored by the two hemispheres may simply be impossible to execute at the same time because both hemispheres must ultimately converge on a single set of neural structures that control the distribution of attention across space. Thus, the phonetic mode of processing favored by the left hemisphere is associated with a very rapid distribution of attention across the three letter stimuli, whereas the mode of processing favored by the right hemisphere is associated with a relatively slow, letter-by-letter, sequential distribution of attention. Hellige (1995) argues that the two hemispheres may be able to work more fully in parallel when they are confronted with stimuli that do not evoke such radically different and incompatible processing modes within each of the hemispheres. Similar incompatibilities may arise in Grimshaw’s (1998) data: subjects who may have manifest complementary lateralization of function when each task was presented alone may not have been able to maintain that complementarity when the two tasks were presented together. Apropos of that, it is interesting to examine the proportion of subjects in her data who had complementarity vs the proportion that did not. Grimshaw recruited 32 rightand 32 non-right-handed participants. Only about 60% of the right-handed subjects manifested complementarity and only about 50% of the nonrighthanders manifested complementarity. These proportions seem close to chance and may suggest that the dichotic-listening assay of hemispheric specialization may not have been adequate or that complementarity cannot be sustained when both kinds of processing compete for output. As Bryden (1986) argued, even though complementarity of hemispheric specialization is more likely to be statistical than causal, and quite independent factors are likely to lateralize cerebral specialization for language vs nonlanguage functions, nonetheless on a statistical basis the probability of complementarity should be between 70 and 80%, at least in right-handers. An entirely different reason why subjects with complementary cerebral specialization did not show less intertask interference than subjects with one hemisphere specialized for both tasks was that Grimshaw’s paradigm was not a bona fide dual-task paradigm. It was a divided-attention task. Subjects were told to attend to one dimension of the stimulus while filtering out the

202

JACQUELINE LIEDERMAN

other dimension of the stimulus. Thus, subjects were consciously attempting to attend to only one input. The following proposed experiment could test whether it was critical to (a) eliminate the identification aspect of the paradigm and (b) ensure that the task is a dual-task rather than a divided-attention paradigm. If subjects were not cued to whether the task would require attention to the tone or the word until after the trial, they would need to attend to both the voice and the word in order to respond. In this proposed paradigm, subjects might hear the word glad in a mad voice, and then a touch-screen display would appear with either the emotion words: MAD, GLAD, SAD, FAD or pictures of facial expressions that were mad, glad, sad, or neutral. The appearance of the words or faces would signal that the word or the emotional tone was the relevant dimension. Subjects would press the screen to indicate which of the four responses is relevant. Reaction times to the appearance of the visual display would be the dependent measure. By this method, the task would be a dual-task paradigm, not a divided-attention paradigm. In addition, the response would be more of a recognition than an identification task, per se. Once again, the prediction would be that individuals with complementary specializations would process the conflicting prosodic/linguistic stimuli faster than those individuals with one hemisphere dominant for both processes. Hellige et al. (1998) examine a different aspect of the bilateral advantage. Their goal is to examine correlates of the bilateral advantage when both hemispheres have equivalent and redundant access to the input compared to unihemispheric nonredundant input. Hellige et al. (1998) interpret the qualitative pattern of errors that occurs during the bihemispheric redundant trials in terms of the pattern of errors that occurs during single-hemisphere stimulation. They find that even though their task (naming of consonant–vowel– consonant syllables) is ostensibly best performed predominantly by the left hemisphere, the pattern of errors during bihemispheric presentation is consistent with right-hemisphere metacontrol. Thus, even though the pattern of errors suggests that the ‘‘wrong’’ hemisphere is dominating during bihemispheric presentation, both hemispheres must be contributing to overall performance because performance in response to redundant bilateral trials exceeds that during unilateral presentations. Hellige et al. (1998) go beyond merely assessing the circumstances under which a bilateral advantage emerges. They attempt to relate unilateral and bihemispheric performance to the size of the corpus callosum. Because ‘‘individual variability far overshadows any other trend in callosal morphology that has thus far been investigated’’ (Byne, Bleier, & Houston, 1988), looking at correlates of variations in callosal size would seem to be instructive. However, searching for such correlates of callosal size assumes that variations in callosal size and function are quantitative. Kinsbourne (1996) has named and endorsed (Yazgan, Wexler, Kinsbourne, Peterson, & Leckman, 1995) this assumption, which he refers to as the ‘‘continuity assumption.’’

INTERHEMISPHERIC COLLABORATION

203

The continuity assumption centers on the belief that there is a functional continuum between the thickest callosum and the thinnest (or absent) callosum. This assumption may not be warranted. The reason is that the differential origin of callosal thinning (or enlargement) may have critical implications for qualitative variations in callosal connectivity. These qualitative variations might include the pattern of origin and/or termination of fibers not only in terms of density, but also in terms of topographic and laminar position (Barbas, 1986), kind of terminal aborization (Innocenti, 1994), degree of collateralization; whether the fiber is homo- or heterotopic, degree of myelination, etc. Taken together, these variations almost certainly lead to qualitative differences that may surpass the implications of the quantitative variations. For example, perinatal disturbances of thalamocortical input are often associated with maintenance of exuberant connections and therefore enlargement of the corpus callosum. Changes in the thalamocortical input can be due to monocular occlusion, enucleation, closure, squint (Innocenti & Frost, 1979; Innocenti, Frost, & Illes, 1985; Lund, Mitchell, & Henry, 1978), or direct damage to thalamocortical afferents (Finlay & Miller, 1993). Similarly, it is known that variations in the prenatal chemical environment, such as levels of sex steroids (Fitch, Berrebi, Cowell, Schrott, & Denenberg, 1990; Fitch, Cowell, Schrott, & Denenberg, 1990) or exposure to alcohol (Riley, Mattson, Sowell, Jernigan, Sobel, & Jones, 1995), have significant effects upon the cross-sectional area of distinct regions of the callosum. Hence ‘‘more’’ may be ‘‘different’’ in a very fundamental way. Interpretation of changes in callosal size is further confounded by the fact that the width of the corpus callosum does not correlate with the number of fibers within it. For example, in rhesus monkeys, there is approximately twofold variation among animals between the density of axons and the midsagittal area of the corpus callosum, and there is no correlation between the area of the corpus callosum and the number of axons (LaMantia & Rakic, 1990). Similarly, Finlay and Miller (1993) found that perinatal anoxia in kittens led to a major increase in the number of callosally projecting neurons. Yet the cross-sectional callosal area did not change, suggesting that the increase in cell number was accompanied by a decrease in myelination or fiber caliber. Therefore, seeking functional correlates of cross-sectional callosal area, per se, may present interpretational difficulties. Hellige et al. (1998) measured: (a) asymmetries within certain cortical regions, (b) the cross-sectional area of the corpus callosum, and (c) various behavioral asymmetries. They measured all three variables, but only compared the relations between a and b and between a and c. It would have been wonderful if they had done a full partial correlation analysis of the midsagittal callosal area, cortical asymmetries, and asymmetries of function. That would have enabled one to examine the relative strength of the contribution of anatomical asymmetry vs callosal size to behavioral asymmetry.

204

JACQUELINE LIEDERMAN

One could argue that it would be hard to correlate the size of a particular cortical region with a particular segment of the corpus callosum because each sector contains fibers from diverse cortical origins. For example, the body and the dorsal splenium of the cat contain visual, somatosensory, limbic, and auditory fibers (Lomber, Payne, & Rosenquist, 1994). Still, given that all three pieces of data were available in each subject, it might nevertheless be informative in that one might expect a path-type of analysis to reveal that the relation between callosal size and asymmetries in functions is driven by the relation between callosal size and cortical asymmetry. The reason for this prediction is that it has been argued from several points of view that cortical asymmetries probably determine callosal development more than callosal development affects cortical asymmetries. Aboitiz, Scheibel, Fischer, and Zaidel (1992) argued this on the grounds that during fetal development, anatomical cerebral asymmetries emerge prior to callosal development (in humans, for example, asymmetries in the Sylvian fissure have been reported as early as 31 weeks of gestation (Chi, Dooling, & Floyd, 1977)). Similarly, Lassonde, Bryden, and Demers (1990) demonstrated that individuals who developed without a corpus callosum (i.e., those with agenesis of the corpus callosum) have normal functional cerebral asymmetry, as assessed by dichotic listening (adjusting for their low IQ scores). Thus, the corpus callosum is not necessary for functional asymmetries to arise. In contrast, there may be a mechanism by which cortical asymmetries determine callosal size. Rosen, in a commentary after Rosen, Galaburda, and Sherman (1990), conjectured that the smaller of the two cortices may determine the number of callosal projections, because there may be some kind of restriction that all callosal pathways have to be reciprocal. In that case, there would be more instances of retraction of the callosal axon and retention of the intrahemispheric ipsilateral projection of a particular cortical neuron on the larger side rather than on the smaller side. What Rosen, Sherman, and Galaburda (1989) have demonstrated is that, in rats, interhemispheric connections differ between asymmetrical and symmetrical brains. As the absolute degree of asymmetry increased between individuals, the number of patches of callosal terminations decreased. There is some very indirect evidence in humans to support Rosen’s conjecture that asymmetric brains have smaller callosa. One group of individuals with asymmetric brains might be those with ‘‘fluctuating asymmetry.’’ Fluctuating asymmetry represents variation from the normal degree of symmetry in physical features that occurs at the population level. An example of a fluctuating asymmetry is differential ear length. At the level of a population trait, the left and right ears are equal in size. Asymmetry in ear length reflects the effects of random environmental stressors on the developmental process. A similar computation can be made for structures when a population directional asymmetry exists (e.g., for the size of the planum temporale). One merely corrects for the directional bias by calculating deviation from the

INTERHEMISPHERIC COLLABORATION

205

population average of asymmetry for each individual. On the basis of Rosen et al.’s (1990) theory, one would predict that (1) the more fluctuating asymmetry an individual manifests on somatic structures, then (2) the more fluctuating asymmetry an individual will manifest in brain structures, and then (3) the smaller will be that individual’s corpus callosum, because on a fine level, wherever there was asymmetry, the smaller cortical region would determine the number of callosal fibers. This is essentially what is reported by Thoma, Yeo, Gangestad, Lewine, Hill, and Orrison (1996), though some of these exact correlations were not computed by these researchers. They reported, based upon 28 normal adult MRI scans, that a summary measure of fluctuating asymmetries of body parts that are symmetric on a population level (feet, ankles, elbows, wrists, hands, and length and width of ears, and palmar ridge patterns) was correlated with a summary measure of brain fluctuating asymmetries (calculated from directional asymmetries for cerebral hemispheres, cortical gray matter, and planum temporale). In addition, the cross-sectional area of the corpus callosum correlated negatively with a composite measure of developmental instability that included fluctuating asymmetries in addition to other measures. (Unfortunately, the authors do not report attempting to correlate corpus callosum cross-sectional area with their measures of body and brain fluctuating asymmetries alone.) In addition, all analyses were done with sex partialled out, but they were not done with total volume of the cerebral hemispheres partialled out. This was potentially an important confounding variable because high developmental instability had marginally smaller overall cerebral hemisphere volumes. Taken together, all of these data suggest that the degree of asymmetry should affect the size of the corpus callosum (more than vice versa), such that the greater the asymmetry between two cortical regions, the smaller the size of the corpus callosum coursing between these two regions. CONCLUSIONS

Despite years of research, there is little agreement as of yet about the role of the corpus callosum in the control of interhemispheric interaction. Though it may seem that structural and functional MRI will open windows of understanding, at this time both of these methods are limited in the ways that were just discussed. Thus, so far, important progress has been made by means of behavioral methods, through well-designed and ingenious paradigms such as those employed by Banich (1998), Grimshaw (1998), and Hellige et al. (1998). The fact that behavioral data are important relates quite well to an anecdote that Phil Bryden wrote for his commentary for the TENNET meeting: Many years ago, I taught a very bright young undergraduate who had planned to major in mathematics but had become interested in human experimental psychology.

206

JACQUELINE LIEDERMAN

He had good ideas and excellent critical ability, and went on to graduate work in psychology. Many years later, I encountered him as a professor in a major institution, where he told me about his most recent work, which was purely physiological in nature. ‘‘What happened to the behavior?’’ I asked. ‘‘I don’t do that any more,’’ he said. ‘It’s too hard.’’

REFERENCES Aboitiz, F., Scheibel, A. B., Fischer, R. S., & Zaidel, E. 1992. Individual differences in brain asymmetries and fiber composition in the human corpus callosum. Brain Research, 598, 154–161. Asbjørnsen, A., & Bryden, M. P. 1997. The attention shift index: A Measure of attentional shift in dichotic listening. Journal of the International Neuropsychological Society, 3, 42. Asbjørnsen, A., Bryden, M. P., & Ofte, S. 1997. Push and pull is all the same? Different attentional manipulations in dichotic listening do not give specific effects. Journal of the International Neuropsychological Society, 3(1), 42. Banich, T. M. 1998. The missing link: The role of interhemispheric interaction in attentional processing. Brain and Cognition, 36, 128–157. Barbas, H. 1986. Pattern in the laminar origin of corticocortical connections. Journal of Comparative Neurology, 252, 415–422. Boucher, R., & Bryden, M. P. 1997. Laterality effects in the processing of melody and timbre. Journal of the International Neuropsychological Society, 3, 43. Bryden, M. P. 1986. The nature of complementary specialization. In F. Lepore´, M. Ptito, & H. H. Jasper (Eds.), Two hemispheres—one brain: Functions of the corpus callosum. New York: A. R. Liss. Pp. 463–469. Bryden, P., & MacRae, L. 1988. Dichotic laterality effects obtained with emotional words. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 1, 171–176. Bulman-Fleming, M. B., & Bryden, M. P. 1994. Simultaneous verbal and affective laterality effects. Neuropsychologia, 32, 787–797. Byne, W., Bleier, R., & Houston, L. 1988. Variations in human corpus callosum do not predict gender: A study using magnetic resonance imaging. Behavioral Neuroscience, 102(2), 222–227. Chi, J. G., Dooling, C. E., & Floyd, H. G. 1977. Left-right asymmetries of temporal speech areas of the human fetus. Archives of Neurology, 34, 346–348. Crebolder, J. M., & Bryden, M. P. 1997. Complementary specialization for verbal and visual tasks. Journal of the International Neuropsychological Society, 3, 50. Dimond, S. 1972. The double brain. Edinburgh: Churchill. Finlay, B. L., & Miller, B. 1993. Regressive events in early cortical maturation: Their significance for the outcome of early brain damage. In A. M. Galaburda (Ed.), Dyslexia and development: Neurobiological aspects of extra-ordinary brains. Cambridge, MA: Harvard Univ. Press. Pp. 1–20. Fitch, H. R., Berrebi, S. A., Cowell, E. P., Schrott, M. L., & Denenberg, V. H. 1990. Corpus callosum: Effects of neonatal hormones on sexual dimorphism in the rat. Brain Research and Behavior, 515, 111–116. Fitch, H. R., Cowell, E. P., Schrott, M. L., & Denenberg, V. H. 1990. Corpus callosum: Demasculinization via perinatal anti-androgen. International Journal of Developmental Neuroscience, 9, 35–38. Grimshaw, G. M. 1998. Integration and interference in the cerebral hemispheres: Relations with hemispheric specialization. Brain and Cognition, 36, 108–127.

INTERHEMISPHERIC COLLABORATION

207

Hellige, J. B. 1995. Coordinating the different processing biases of the left and right cerebral hemispheres. In F. L. Kitterle (Ed.), Hemispheric communication: Mechanisms and models. Hillsdale, NJ: Erlbaum. Pp. 347–362. Hellige, J. B., Taylor, B. K., Lesmes, L., & Peterson, S. 1998. Relationships between brain morphology and behavioral measures of hemispheric asymmetry and interhemispheric interaction. Brain and Cognition, 36, 158–192. Houzel, J., Milleret, C., & Buisseret, P. 1992. Early monocular deprivation enlarges widely the area of visual transcallosal responses in A17 and A18 of the cat. European Journal of Neuroscience, 5(Suppl.), 4176. Innocenti, G. 1994. Some new trends in the study of the corpus callosum. Behavioral and Brain Research, 64, 1–8. Innocenti, G. M., & Frost, D. O. 1979. Effects of visual experience on the maturation of the efferent system to the corpus callosum. Nature, 280, 231–233. Innocenti, G. M., Frost, D. O., & Illes, J. 1985. Maturation of visual callosal connections in visually deprived kittens: A challenging critical period. Journal of Neuroscience, 5(2), 255–267. Just, M. A., Carpenter, P. A., Keller, T. A., Eddy, N. F., & Thulborn, K. R. 1996. Brain activation modulated by sentence comprehension. Science, 274, 114–116. Keillor, J. M., Grimshaw, G. M., & Bryden, M. P. 1997. Interhemispheric interaction in bilateral lexical decision. Journal of the International Neuropsychological Society, 3(1), 50. Kinsbourne, M. 1982. Hemispheric specialization and the growth of human understanding. American Psychologist, 37, 411–420. Kinsbourne, M. 1987. The material basis of mind. In L. Vaina (Ed.) Matters of intelligence. Boston: Reidel. Pp. 407–426. Kinsbourne, M. 1996. Challenges to the information flow model: The attention factor. NATO Advanced Study Institute. The role of the human corpus callosum in sensory motor integration: Anatomy, physiology, and behavior; individual differences and clinical applications. Tuscany, Italy. Kinsbourne, M., & Hicks, R. E. 1978. Functional cerebral space: A model for overflow, transfer and interference effects in human performance: A tutorial review. In J. Requin (Ed.), Attention and performance VII. Hillsdale, NJ: Erlbaum. Pp. 345–362. LaMantia, A. S., & Rakic, P. 1990. Axon overproduction and elimination in the corpus callosum of the developing rhesus monkey. Journal of Neuroscience, 10, 2156–2175. Lassonde, M., Bryden, M. P., & Demers, P. 1990. The corpus callosum and cerebral speech lateralization. Brain and Language, 38, 195–206. Ley, R. G., & Bryden, M. P. 1982. A dissociation of right and left hemispheric effects for recognizing emotional tone and verbal content. Brain and Cognition, 1, 3–9. Liederman, J. 1986. Subtraction in addition to addition: Dual task performance improves when tasks are presented to separate hemispheres. Journal of Clinical and Experimental Neuropsychology, 8, 486–502. Liederman, J. 1995. A reinterpretation of split-brain syndrome: Implications for the function of corticocortical fibers. In R. Davidson & K. Hughdahl (Eds.), Brain Asymmetry. Cambridge, MA: M.I.T. Press. Pp. 451–490. Liederman, J., & Meehan, P. 1986. When is between-hemisphere division of labor advantageous? Neuropsychologia, 24(6), 863–874. Liederman, J., Merola, J. L., & Hoffman, C. 1986. Longitudinal data indicate that hemispheric independence increases during early adolescence. Developmental Neuropsychology, 2(3), 183–201.

208

JACQUELINE LIEDERMAN

Liederman, J., Sohn, Y., Thome, M., & Palomo, D. 1995. Factors which affect the degree of interhemispheric interference. Paper presented at the Society for Cognitive Neuroscience, San Francisco, CA. Lomber, S., Payne, B. R., & Rosenquist, A. C. 1994. The spatial relationship between the cerebral cortex and fiber trajectory through the corpus callosum of the cat. Behavioural Brain Research, 64, 25–35. Lund, R. D., Mitchell, D. E., & Henry, G. H. 1978. Squint-induced modification of callosal connections in cats. Brain Research, 144, 169–172. Merola, J. L., & Liederman, J. 1985. Developmental changes in hemispheric independence. Child Development, 56, 1184–1194. Merola, J. M., & Liederman, J. 1987. Developmental vs. individual differences in the ability of the hemispheres to operate independently. International Journal of Neuroscience, 35, 195–204. Merola, J. L., & Liederman, J. 1990. The effect of task difficulty upon the extent to which performance benefits from between-hemisphere division of inputs. Journal of Neuroscience, 51, 35–44. Nikelski, J., Grimshaw, G. M., Bryden, M. P., & Cocivera, T. 1997. Relations between lateralization of semantic and phonological processes. Journal of the International Neuropsychological Society, 3, 51. Norman, W. D., Jeeves, M. A., Milne, A., & Ludwig, T. 1992. Hemispheric interactions: The bilateral advantage and task difficulty. Cortex, 28, 623–642. Obrzut, J. E., Bryden, M. P., Lange, P., & Bulman-Fleming, B. 1997. Concurrent left-hemisphere-verbal and right-hemisphere-emotional processing in children: Dichotic laterality effects. Journal of the International Neuropsychological Society, 3, 50. Pashler, H., & O’Brien, S. 1993. Dual-task interference and the cerebral hemispheres. Journal of Experimental Psychology: Human Perception and Performance, 19, 315–330. Rapp, B., & McCloskey, M. 1997. Brain activation and sentence comprehension. Science, 27, 912–913. Riley, E., Mattson, S., Sowell, E., Jernigan, T., Sobel, D., & Jones, K. 1995. Abnormalities of the corpus callosum in children prenatally exposed to alcohol. Alcoholism: Clinical and Experimental Research, 19, 1198–1202. Ringo, J., Doty, R., Demeter, S., & Simard, P. 1994. Time is of the essence: A conjecture that hemispheric specialization arises from interhemispheric conduction delay. Cerebral Cortex, 4, 331–343. Rosen, G., Galaburda, A., & Sherman, G. 1990. The ontogeny of anatomic asymmetry, constraints derived from basic mechanisms. In A. Scheibel & A. Wechsler (Eds.), Neurobiology of higher cognitive function. New York: Guilford. Pp. 215–238. Rosen, G. D., Sherman, G. F., & Galaburda, A. M. 1989. Interhemispheric connections differ between symmetrical and asymmetrical brain regions. Neuroscience, 33, 525–533. Sohn, Y. S., Liederman, J., & Reinitz, M. T. 1996. Division of inputs between the hemispheres eliminates illusory conjunctions: Evidence of hemispheric independence. Neuropsychologia, 34, 1057–1068. Thoma, R., Yeo, R., Gangestad, S., Lewine, J., Hill, D., & Orrison, W. W. 1996. Developmental instability (DI) and the brain: DI predicts hemispheric asymmetries. Presented at Cognitive Neuroscience, San Francisco, CA. Treisman, A., & Schmidt, H. 1982. Illusory conjunctions in the perception of objects. Cognitive Psychology, 14, 107–141. Yazgan, M. Y., Wexler, B. E., Kinsbourne, M., Peterson, B., & Leckman, J. F. 1995. Functional significance of individual variations in callosal area. Neuropsychology, 33(6), 769– 779.